Pulse laser-induced generation of cluster codes from metal nanoparticles for immunoassay applications

In this work, we have developed an assay for the detection of proteins by functionalized nanomaterials coupled with laser-induced desorption/ionization mass spectrometry (LDI-MS) by monitoring the generation of metal cluster ions. We achieved selective detection of three proteins [thrombin, vascular endothelial growth factor-A165 (VEGF-A165), and platelet-derived growth factor-BB (PDGF-BB)] by modifying nanoparticles (NPs) of three different metals (Au, Ag, and Pt) with the corresponding aptamer or antibody in one assay. The Au, Ag, and Pt acted as metal bio-codes for the analysis of thrombin, VEGF-A165, and PDGF-BB, respectively, and a microporous cellulose acetate membrane (CAM) served as a medium for an in situ separation of target protein-bound and -unbound NPs. The functionalized metal nanoparticles bound to their specific proteins were subjected to LDI-MS on the CAM. The functional nanoparticles/CAM system can function as a signal transducer and amplifier by transforming the protein concentration in...

In this work, we have developed an assay for the detection of proteins by functionalized nanomaterials coupled with laser-induced desorption/ionization mass spectrometry (LDI-MS) by monitoring the generation of metal cluster ions. We achieved selective detection of three proteins [thrombin, vascular endothelial growth factor-A 165 (VEGF-A 165 ), and platelet-derived growth factor-BB (PDGF-BB)] by modifying nanoparticles (NPs) of three different metals (Au, Ag, and Pt) with the corresponding aptamer or antibody in one assay. The Au, Ag, and Pt acted as metal bio-codes for the analysis of thrombin, VEGF-A 165 , and PDGF-BB, respectively, and a microporous cellulose acetate membrane (CAM) served as a medium for an in situ separation of target protein-bound and -unbound NPs. The functionalized metal nanoparticles bound to their specific proteins were subjected to LDI-MS on the CAM. The functional nanoparticles/CAM system can function as a signal transducer and amplifier by transforming the protein concentration into an intense metal cluster ion signal during LDI-MS analysis. This system can selectively detect proteins at picomolar concentrations. Most importantly, the system has great potential for the detection of multiple proteins without any pre-concentration, separation, or purification process because LDI-MS coupled with CAM effectively removes all signals except for those from the metal cluster ions. © 2017 Author(s). All  The greatest challenge in the detection of specific proteins or tumor markers for the diagnosis of cancer is their low concentrations in human plasma. 1,2 Interferences due to other proteins with similar properties also cause difficulties in their selective detection by conventional methods. 2 Therefore, to achieve high selectivity for the identification of specific proteins in biological fluids, such as plasma, immunoassays based on aptamer (Apt)-protein-specific and antigen-antibody (Ab)-specific interactions are widely used in both clinical and medical research. 3 Currently, the enzyme-linked immunosorbent assay (ELISA) has demonstrated reasonable sensitivity and specificity; however, it fails in the analysis of multiple proteins in a single well, which limits its application. 4 Matrix-assisted laser-induced desorption/ionization (MALDI) time-of-flight mass spectrometry (MS) is an effective tool for identifying biomacromolecules such as proteins, nucleic acids, and polysaccharides with great selectivity. 5 However, non-uniform matrix-analyte co-crystallization and interference signals from organic matrices significantly decrease the reproducibility and sensitivity of this method. 6  spectrometry (SALDI-MS) using nanomaterials, such as a matrix (substrate), can improve reproducibility and reduce matrix interference. 7-10 SALDI-MS has been successfully used in the analysis of a wide variety of analytes such as proteins, DNA, microbes, and tumor cells. However, in the case of multiple-target analysis and quantitation by SALDI-MS, the fragmentation of several analytes, background molecules, and their adducts is unpredictable. Thus, the comprehensive detection of complex samples such as plasma containing complicated proteins, small molecules, and salts, is difficult. To resolve the interference from background proteins and the unpredictable fragmentation of analytes encountered in SALDI-MS, we have developed a simple immunoassay that exhibits significant potential for the simultaneous detection of different proteins in a single analysis by monitoring the cluster ion signals generated from the metal nanoparticles (NPs) under pulse laser irradiation. In this work, instead of observing the various intact or fragmented protein ions, we measure the metal-codes (specific metal cluster ions) from the NPs itself for the quantitative detection of the three analytes: thrombin, vascular endothelial growth factor-A 165 (VEGF-A 165 ), and plateletderived growth factor-BB (PDGF-BB). These proteins play critical roles in angiogenesis and tumor progression. [11][12][13] Thrombin promotes angiogenesis by activating PAR1 receptors in platelet and endothelial cells. 11 VEGF and PDGF are signal proteins, which are highly expressed by tumor cells to stimulate tumor angiogenesis and vascular remodeling by binding to specific receptors on endothelial cells. 12,13 Therefore, the determination of the concentration of these three cytokines in plasma and in tumor environments is very important for the diagnosis of tumor growth and metastasis. [11][12][13] The functionalization of different metal NPs with their respective aptamers or antibodies enables specific targeting of thrombin, VEGF-A 165 , and PDGF-BB ( Fig. 1). Here, three metal (Au, Ag, and Pt) NPs are used as mass tags for the proteins, rather than a SALDI matrix, to enhance the ionization of the analyte molecules. The metal NPs absorb pulsed laser energy and undergo photothermal evaporation and/or Coulombic explosion, which produces a substantial amount of cluster ions useful for signal amplification. [14][15][16][17] We used cellulose acetate membrane (CAM), which can be directly mounted onto a plate for LDI-MS analysis, to serve as a medium (substrate) to separate the antibody (Ab)-or aptamer (Apt)-modified NPs and their conjugates formed with their targeting proteins in situ. This assay does not require any additional processing steps such as separation, preconcentration, or washing. Because of the increased particle weight or decreased affinity towards CAM upon interaction with the target proteins, target-bound nanoparticles penetrate deeper into the CAM. indicate that the as-prepared Au NPs, Ag NPs, and Pt NPs have average particle sizes of ∼13, ∼26, and ∼24 nm, respectively (Fig. S1, supplementary material). The UV-Vis absorption and X-ray diffraction (XRD) spectra show the surface plasmon resonance band and crystal structures, respectively, which further confirm the formation of corresponding metal nanoparticles (Fig. S2, supplementary material). From dynamic light scattering (DLS) measurements (Fig. S3, supplementary material), the increased hydrodynamic size (∼20 nm) of the NPs observed after aptamer-or antibody-modification supports that the aptamer or antibody ligands are anchored on the NPs' surfaces. To demonstrate our detection strategy, the fabrication of a probe [aptamer-modified gold nanoparticles (Apt-Au NPs)] for the detection of thrombin is described in detail. To achieve specificity, Au NPs were functionalized with two types of thiol-modified thrombin-binding aptamers (TBAs): a 15-base-long aptamer (TBA 15 ), which interacts with exosite I of thrombin, and a 29-base-long aptamer (TBA 29 ), which binds with exosite II of thrombin. 18 20 The average number of TBA molecules on the surface of Au NPs was determined to be 70 TBA molecules per Au NP by OliGreen (OG) labeling of TBA in the supernatant after centrifugation. Pulsed laser irradiations have been employed for the reshaping and fragmentation of metal NPs. [14][15][16][17] The pulsed laser irradiation of metal NPs is accompanied by the production of ultra-small cluster ions. The intensity and size of the formed cluster ions are highly dependent on the laser power density, pulse width, surface properties of metal NPs, and surrounding pressure. Au NPs (∼13 nm) irradiated with a pulse laser (Nd:YAG, 355 nm) of sufficient power density tend to undergo fragmentation via photothermal evaporation and/or Coulombic explosion, producing Au cluster ions that can be detected using a mass analyzer. 17 The influence of laser power and possible physical and chemical phenomena involved in the fragmentation of Au NPs has been discussed in detail in our previous reports. [21][22][23] Here, the signal intensities of the Au cluster ions ([Au n ] + ; 1 ≤ n ≤ 3) were monitored at an applied laser power density of 3.  (Fig. 2(a)), in contrast to the LDI-MS analysis without CAM (Fig. S4). The decrease in the zeta potential of TBA 15 /TBA 29 -Au NPs (1 nM) from 34.6 ± 1.3 mV to 13.8 ± 0.9 mV after reaction with thrombin (100 nM) reveals that thrombin molecules were bound to the surfaces of the nanoparticles. The scanning electron microscopy (SEM) images of 5 µl TBA 15 /TBA 29 -Au NPs (1 nM) in the absence and presence of 100 nM thrombin on CAM are shown in Fig. S6 (supplementary material). The average number of NPs per 100 µm 2 on the surface of membranes was counted and found to be approximately a factor of 2 lower for TBA 15 /TBA 29 -Au NPs treated with 100 nM thrombin.
CAM is made up of a negatively charged hydrophilic porous network (porosity approximately 70%) of cellulose diacetate and triacetate fibers. TBA 15 /TBA 29 -Au NPs are uniformly spread on the membrane and penetrated into the pores by both gravity and capillary action. 24 The movement of the particles through the pores of the membrane along with the solution is dependent on the size and surface properties of the nanoparticles. As the thrombin concentration increased, the TBA 15 /TBA 29 -Au NPs bound more thrombin. As a result, the heavier TBA 15  penetrated faster into the CAM. Therefore, as the concentration of thrombin increased, the number of TBA 15 /TBA 29 -Au NPs on the surface of the CAM decreased. Additionally, thrombin bound to the surfaces of TBA 15 /TBA 29 -Au NPs may compromise the interaction between the highly dense TBA ligands and the cellulose acetate fiber and may also contribute to its faster migration (penetration). Substituting CAM with nitrocellulose membrane (NCM) and mixed cellulose ester membrane [MCEM; composed of cellulose nitrate (80%) and cellulose acetate (20%)] produced similar results for the sensing of thrombin (Fig. S7, supplementary material), whereas a positively charged nylon membrane (N + M) did not work in this system. The porous N + M has a high binding capacity for nucleic acids-as high as to 600 µg cm 2 -because of its high number of positively charged quaternary ammonium groups. Thus, the TBA 15 /TBA 29 -Au NPs cannot easily penetrate into the pores because of the strong electrostatic interaction between the N + M fiber and the nanoparticles.
Compared to the noisy spectra typically obtained when using nanoparticle-assisted LDI-MS 7-10 or our previously described TBA 15 /TBA 29 -Au NPs ( Fig. S4(a)), the MS spectra of the TBA 15 /TBA 29 -Au NPs/CAM system are very clean ( Fig. 2(a)), presumably because the negatively charged porous cellulose acetate fiber effectively binds the interfering cationic molecules produced under LDI. 21 We cannot rule out the possibility that the clean MS spectra are due to small molecules and salt ions depositing not on the top but rather on the bottom of the CAM. In addition, the relative standard deviation (RSD) of the MS signals of [Au 1 ] + obtained from the same TBA 15 /TBA 29 -Au NPs/CAM substrate, collected from 50 different mass spectra, was less than 10%, revealing a high homogeneity of nanoparticle distribution on the CAM. In our previous study, we have demonstrated that the microporous membrane is an ideal substrate for homogenous deposition of Au NPs. 24 We conducted control experiments under similar conditions; however, instead used a random oligonucleotide (base number same as TBA 29 )-capped Au NPs for the analysis of thrombin. As expected, the addition of thrombin (10 nM) did not induce any substantial changes in the signals of the [Au n ] + cluster ions (Fig. S8, supplementary material). We also evaluated the selectivity of the TBA 15 /TBA 29 -Au NPs/CAM as an LDI-MS substrate for the analysis of various proteins (10 nM for thrombin, 1.0 µM for each of the other proteins) in the presence of BSA (100 µM). A plot of the relative signal intensity changes of the [Au 1-3 ] + revealed that this system was highly selective (1000-fold or more) toward thrombin over the other proteins (Fig. S9, supplementary material). Our TBA 15 /TBA 29 -Au NPs/CAM coupled with LDI-MS allows for detection of thrombin at concentrations as low as 10 pM (Fig. 2(b)) in the presence of 100 µM BSA (i.e., a 1 × 10 7 -fold higher concentration), further demonstrating the system's high selectivity. The high selectivity of the TBA 15 /TBA 29 -Au NPs/CAM probe is due to the high specificity and strong binding between TBAs and thrombin. Moreover, background proteins are bound by CAM, resulting in low interferences. In comparison with other methods (Table S1, supplementary material), our sensing platform for thrombin is relatively simple, rapid, and sensitive. Note that most other methods require tedious labeling of nanoparticles and complicated separation, preconcentration, and/or washing processes during sensing.
We further used our LDI-MS-based sensing system to detect other proangiogenic factors. Ag NPs and Pt NPs were modified with VEGF-A 165 antibody (Ab VEGF ) and PDGF-BB antibody (Ab PDGF ) to form functional Ab VEGF -Ag NPs and Ab PDGF -Pt NPs for the detection of VEGF-A 165 and PDGF-BB, respectively (details regarding the preparation of antibody-Ag or -Pt NPs are included in the experimental section of the supplementary material). Flocculation assay studies suggest that the average numbers of antibodies modified per Ag NP and Pt NP are approximately 70 and 80 molecules (data not shown), respectively (see details in supplementary material). 25 Both the as-prepared Ab VEGF -Ag NPs and Ab PDGF -Pt NPs are stable (no aggregation) in a physiological solution containing 100 µM BSA. Similarly, Ab VEGF -Ag NPs/CAM and Ab PDGF -Pt NPs/CAM coupled with LDI-MS exhibit high sensitivity [limits of detection (LODs) of approximately 5 and 50 pM (based on a signal-tonoise (S/N) ratio of 3), respectively] and selectivity (>1000-fold relative to other proteins) toward their target proteins, VEGF-A 165 , and PDGF-BB ( Fig. 3 and Fig. 4), respectively. The homodimeric characteristic of PDGF-BB induces significant crosslinking aggregation of Ab PDGF -Pt NPs (Fig. S10, supplementary material). 26 As a result, the aggregated Pt NPs cannot penetrate into the porous CAM and instead on the surface of the CAM (Fig. S11, supplementary material). Therefore, the mass signal of [Pt 1 ] + ions increases upon increasing the concentration of PDGF-BB ( Fig. 4(a)). The LOD of the detection of PDGF-BB (50 pM) is relatively higher than that of VEGF-A 165 (5 pM) and thrombin (10 pM), mainly because it fails to induce large degree of aggregation of Ab PDGF -Pt NPs, at very low concentrations.
By measuring their respective metallic cluster ions ([Au 1 ] + or [Ag 1 ] + ), we also demonstrated that TBA 15 /TBA 29 -Au NP and Ab VEGF -Ag NP probes enable the selective simultaneous detection of thrombin and VEGF-A 165 . As shown in Fig. 5, the intensities of the signals of the metallic cluster ions ([Au 1 ] + and [Ag 1 ] + ) decreased when only their corresponding protein was present. This result demonstrates that the system we developed is superior to ELISA for the detection of multiple proteins in a single assay. We also attempted to use the three NP probes for simultaneously detecting thrombin, VEGF-A 165 , and PDGF-BB (Fig. S12, supplementary material). Unfortunately, the TBA 15 /TBA 29 -Au NP and Ab VEGF -Ag NP probes exhibit nonspecific binding to PDGF-BB. TBA 15 /TBA 29 -Au NPs interact with PDGF-BB probably due to the strong electrostatic interaction between the highly dense, negatively charged aptamer ligands on Au NPs, and the positively charged PDGF-BB (isoelectric point ∼9.8). On the other hand, similarity in protein structure (subfamily) of VEGF and PDGF may also result in cross talk of their antibody-modified NPs. 27 To test the practicality of the newly developed sensing system, we analyzed proangiogenic factors (thrombin, VEGF-A 165 , and PDGF-BB) in human plasma. The relative signals of the metal cluster ions of [Ag 1 ] + and [Pt 1 ] + increased linearly with increasing concentrations of the spiked VEGF-A 165 and PDGF-BB, respectively (Figs. 6(b) and 6(c)). The LODs (S/N = 3) for VEGF-A 165 and PDGF-BB in plasma are approximately 25 and 200 pM, respectively. The recoveries for the spiked VEGF-A 165 and PDGF-BB are determined to be 94%-106% and 95%-109%, respectively. Even though our proposed approach appears to be applicable to the practical screening of proangiogenic factors in complex biological samples, our assay unfortunately failed to detect thrombin in the plasma ( Fig. 6(a)), primarily because of the nonspecific binding between basic proteins in the plasma and the original surface properties of the Au NPs being modified by TBA ligand. Antibody-modified Au NPs may function as an alternative to TBA in the future to improve the specificity for the detection of thrombin in plasma.
In summary, we have demonstrated a simple nanomaterial-assisted method using LDI-MS coupled with CAM-mediated separation for the detection of proteins. The CAM employed in this study not only acts as a separation matrix but also suppresses the fragmentation of ligands functionalized on the metal NPs and target proteins, which leads to a clean mass spectrum, especially in the low-molecular-weight region. Monitoring of the MS signal of metal cluster ions from metal NPs in LDI-MS provides greater sensitivity relative to that of intact proteins or surface ligands because of the poor ionization efficiency and easy fragmentation of proteins and surface ligands. [28][29][30][31][32] Furthermore, monitoring the changes in the cluster ions' intensity of metal bio-codes enables the quantification of different proteins using a single assay. We hope the principles applied in this work offer a new direction for the development of multiplex immunoassays.
See supplementary material for additional information (experimental section of materials and LDI-MS, Table S1, and Figures S1-S12) which is noted in the text. This material is available free of charge via the Internet at http://dx.doi.org/XXXX.